A Grammar of Tamashek (Tuareg of Mali)

4.1 Noun morphology 189

e. minor types

e-merwas i-merwis-aen (A-grm) 'sickly person'

[Sg e-merwas, PI i-merwas-sen (T-ka)]

α-laggas i-laggus-asn (A-grm) 'brother-in-law'

[i-laws-cm (T-ka)]

maejraed majrid-aen 'speech, speaking'

Of the "minor types" in (164.e), only 'speech' occurs in my Timbuktu-area
data in the forms shown. The PI majrid-aen suggests VblN status, but both Sg
and PI lack vocalic prefixes and the Sg/Pl alternations seems isolated and
irregular.
The VblN forms shown in (164.a-d), with a in the final Sg syllable, have
variants with ο (§8.6.1.4). The latter variant retains its α in the PI, as in Sg
a-lawlaw 'towering', PI i-lawlaw-aen.
Formally, we can account for the lengthening in (164) as the effect of an
ablaut component χ-f (i.e., lengthening of stem-final V). The combination of
the length component and the melody applying to the whole stem is i
rather than u in this case.

(165) V-Lengthening (Verbal Nouns)

In VblN's of prefixally derived verbs, and of heavy underived

stems, if the Sg ends in ...CaC, the suffixal PI includes an ablaut

component χ-f whose effect is to lengthen a short V (always

schwa) to a full V in the final stem syllable. Applied to a (i.e. short

vowel with <H> melody), χ-f produces i (except as noted below).

The lengthening applies only to the simple form of the VblN with schwa in
the final Sg syllable, not to the variant forms with α replacing this schwa in
both Sg and PI: Sg a-lawlaw 'towering', PI i-lawlaw-aen, etc. On occasion I
have elicited Sg/Pl pairs with schwa in the Sg and α in the PI, particularly from
my A-grm informant, who gave e.g. Sg α-s-aksal 'shortening' (causative
VblN) and PI a-s-aksal-aen. Perhaps for some speakers the α is more common
in the PI than in the Sg, the effect being at least the appearance of a
morphological rule lengthening a to a. However, my impression is that these
"pairs" may have involved mixing of the simple and α versions of the (Sg and
PI) VblN's.
When the verb stem contains a medial u, the features [+rounded] and
[+back] spread from it to the lengthened V of the stem-final V of the PI VblN,
resulting in another u instead of the expected i. This works for T-ka and most
other dialects investigated. However, the A-grm dialect fails to apply
u-Spreading, and therefore allows the medial u and the final-syllable i to co-
exist. Some examples from T-ka area are in (166).